Those worried about the quality of an AI chatbot’s answers should avoid a Brussels-approved ChatGPT, as should those concerned about privacy.
The EU Commission has recently declared that it would like to censor more American speech and/or loot more U.S. high tech.
Not that they put it that way.
Instead, there was this on X:
We have designated ChatGPT as a Very Large Online Search Engine and Reddit and Roblox as Very Large Online Platforms under the Digital Services Act. They now have four months to comply with additional DSA obligations.
In an article for National Review in 2024, I set out a brief explanation of what these designations meant:
The DSA imposes a wide range of obligations on online-service providers if they offer their services in the EU. These increase substantially in the case of companies that have more than 45 million users a month there and that Brussels has designated as either a very large online search engine (VLOSE) or a very large online platform (VLOP).
A VLOP (such as Reddit or Roblox are now set to be) is, as I explained,
required, among many other obligations, to undertake an annual assessment of “systemic” risks arising out of, to oversimplify, the way its operations are set up and the use that is made of its services. Some risks are obvious (dissemination of illegal content), but others are extraordinarily broad (“any actual or foreseeable negative effects on civic discourse and electoral processes”). The VLOP must then explain how it “mitigates” those risks. It is clearly envisaged that the appropriate response to “illegal hate speech” is to remove it, but the overall requirement is that mitigation should be “reasonable, proportionate, and effective.” In the hands of an aggressive regulator, that could mean anything.
Fines for breaches of the DSA can be up to 6 percent of global revenue.
A VLOSE (such as ChatGPT is now meant to be) is subject to the same rules, but in practice (and to oversimplify), those rules operate in a different way from those covering a VLOP. An online search engine is not (except indirectly) an internet platform where third parties can express themselves.
To Brussels:
ChatGPT is an Artificial Intelligence (AI) system that can engage with and respond to users’ prompts and queries, including by searching the web. Hence, ChatGPT is a hybrid service that qualifies as an online search engine under the DSA.
As a VLOSE, ChatGPT (and the answers it gives or opinions it passes on) will be subject to Brussels’s scrutiny. As the commission explains, ChatGPT will be required
to comply with the additional DSA obligations for VLOPs and VLOSEs, such as assessing and mitigating the systemic risks stemming from their service and algorithmic systems related to the dissemination of illegal content, the negative effects on minors, users’ physical and mental well-being, fundamental rights, electoral processes and public security.
In other words, when catering to users in the EU, ChatGPT will have to do its best to ensure that the answers it gives will not on a systemic basis include anything that the EU Commission could consider to have (among other no-no’s) “negative effects” on “fundamental rights [or] electoral processes,” questions better suited for democratic debate than regulatory procedure. Then again, the EU, an institution designed to evolve away from democracy, would not be expected to agree with that.
Note that phrase “on a systemic basis.” ChatGPT would not be expected to ensure that its every answer avoided falling afoul of the rules. But in practice its programmers would be expected to minimize the chance that the bot gave out a reply that displeased Brussels’s regulators in any review. Then there’s the fact that the EU Commission regards American high tech as a bank to be looted. Given the enormous potential fines involved, ChatGPT will err on the side of caution.
And there is something else. The designation of ChatGPT as a VLOSE would mean that some users’ interactions would (logically) be subject to inspection by Brussels or its proxies as part of the maintenance of a compliance regime. That does not mean that every query would be scrutinized (very, very few would), but it is hard to see how a compliance regime can be enforced without some inspections of ChatGPT’s Q&A. Those worried about the quality of the answers from an AI chatbot should avoid a Brussels-approved ChatGPT, as should those concerned about privacy. No U.S. government department should use it.
This Brussels diktat is censorship, if by sleight of hand, made more pernicious by the way that the nature of AI makes it more difficult for the layperson to verify errors and omissions in the material with which he or she is presented.
Oh yes, there’s more.
As I wrote back in 2024:
The Digital Services Act is not meant to criminalize any new categories of speech. What is illegal under the law of an individual EU member-state or under EU law will remain illegal. Any amendments to legislation in that area will be left to national parliaments or to the EU’s legislative process. However, the DSA’s broad language could easily be used to impose de facto censorship on all sorts of theoretically legal speech, in the interest of preventing “harms” that exist only in the progressive imagination and that are hinted at in, among other places, the law’s preamble [its recitals], but also elsewhere.
And lurking within the DSA’s Recital 84, there is this (my emphasis added):
When assessing the systemic risks identified in this Regulation, those providers should also focus on the information which is not illegal, but contributes to the systemic risks identified in this Regulation.
The DSA recitals themselves are not legally binding, but they are designed to assist in interpreting the law, a distinction all too often without a difference.